Lost Cities

Lost Cities is a Javascript recreation of the popular card game "Lost Cities", both with other players and bots trained on a deep learning model

Installation

1. Pull or download this repo
2. ensure node is installed
3. add the following packages (for console interaction)

• "@colors/colors": "^1.5.0",
• "@tensorflow/tfjs": "^4.8.0",
• "cli-table3": "^0.6.3"

Usage

Interacting with logic components

• PvP through console: node main.js
• PvAi through console: node aiMain.js
• testing: npm test

Interacting with UI components

• cd lost-cities-react
• npm start

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

MIT

To-Do Lists

1. AIModel
   • Initialize game state, updates it after every single move/action (both players)
   • Initialize Q value, updates it after every single move/action (AI only)
   • Come up with all possible states that AI may encounter
   • Create a reward function (Positive rewards for favorable moves or winning the game and vice versa; rewards value can be the same for all favorable moves and vice versa)
   • Create an action selection function (determined by the highest Q value in the Q table)
   • Constructa neural network
     • First layer/ Input layer: Uses for state obervations
     • Second Layer and Third Layer: Based on the policy and Training Loop
     • Fourth Layer / Output Layer: Q values table
2. Convert actions into valid strings based on the Q values and action index
3. Make sure reward function only return positive rewards for favorable moves and vice versa
   • Bug: Cause the game call play/draw function twice there will be an extra card missing after each turn
4. Perhaps train the AI before the game